ABSTRACT
In research on immigration, a lot of attention is devoted to the intergenerational socio-economic mobility of second and third generation immigrants, but little is known about the class mobility of first generation migrants. We address this issue with the use of a representative survey on middle-aged and elderly immigrants in France. A comparison between the socio-economic status of each male respondent during his childhood and at the time of the interview reveals some variation in class mobility across countries and regions of origin. At the same time, our analysis of the main determinants of the socio-economic status of the respondents at the end of their careers highlights a strong influence of the socio-economic environment during their childhoods and gives support to our main hypothesis of a dominance effect.
1. Introduction
Since one of the main motives for migration is the improvement of the standard of living of migrants, it is not surprising that much of the immigration literature tends to associate spatial mobility with the relative financial mobility of individual migrants (Chiswick 1999). Socio-economic or status mobility in the context of migration, on the other hand, has typically been explored either from the point of view of second (or higher) generation migrants compared to their parents (Silberman and Fournier 1999; Gang and Zimmermann 2000; Birnbaum and Werquin 2004; Meurs et al. 2005), or by looking at the evolution of the professional development of first generation migrants along their lifetime career paths. The main message that comes out of this set of studies is that, in general, migrants tend to experience some status degrading in their first jobs in the host society, but progress towards higher socio-economic positions over their lifetimes and especially across generations (Chiswick 1979; Dayan et al.1997; Gang and Zimmermann 2000; Chiswick et al.2005; Recchi 2006).
Little is known about the intergenerational class mobility of first generation migrants, due to both data limitations and the perception that immigration entails a discontinuity or separation of individuals from their community and class of origin (Tyree et al.1979). To the best of our knowledge, only Yaish (2002) compares the socio-economic status of first-generation migrants in Israel with the socio-economic status of their parents in the countries of origin. The main focus of his paper is the impact of the absolute and relative mobility of migrants on the class structure of the society in the destination country. No attention is given to the impact of such intergenerational mobility on individual migrants.
We believe that this highlights an important shortcoming of the literature that tries to evaluate the performance and welfare of migrants and test the hypothesis that social family origin counts also for first generation migrants in determining social mobility processes. We do that by first comparing the socio-economic status of migrants in France who are at the end of their careers with the socio-economic status of their parents in the country of origin, and then exploring the impact of the latter on success in the country of destination. In doing so, we look for support of a popular hypothesis drawn from the sociology of social mobility, which shows evidence of ‘reproduction’ of social positions from fathers to sons (Bourdieu and Passeron 1972; Boudon 1977, 1984). Our results indicate that although the overall mobility patterns differ across countries, and especially continents, of origin, there is a strong influence of the socio-economic environment during the childhoods of migrants in their countries of origin on their socio-economic status towards the end of their life spans. This influence holds across ethnic and national origins. It gives support to our main hypothesis of a dominance effect (Boudon 1977, 1984). Thus, contrary to stylized beliefs, migration does not entail a complete discontinuity or separation of individuals from their community or class of origin. Instead, class of origin has evident implications for the class mobility of individuals even across borders.
The rest of the paper is organized as follows. In the next section we lay out the conceptual framework and empirical specification. Section 3 describes the data. Our regression results are highlighted in section 4 and section 5 concludes.
2. Conceptual framework and empirical specification
2.1. Conceptual framework
According to Lewin-Epstein (2007), the welfare of migrants includes several dimensions of social quality, such as social empowerment, social cohesion and social inclusion. More generally, the literature on life satisfaction shows that it has multiple dimensions, including satisfaction with employement, standard of living, family and social network (De Jong et al. 2002; Neto 2005), as well as satisfaction with physical health (Silveira et al.2002). These dimensions, while correlated with the social position, go beyond the question of social mobility. At the same time, we believe that social mobility is part of the social quality aspect of the migrant's welfare and it is therefore important to deepen the analysis of the migrant's social mobility in the framework of the sociology of migration. In particular, we would like to test the hypothesis of a reproduction of social status from parent to child, or a dominance effect of social origin on a person's social status during adulthood, which is closely related to the theory of dominance de milieu or dominance effect.
This hypothesis bears its origin from the observation that among individuals with the same level of education, but different social backgrounds, those coming from higher strata of the society have a greater probability of achieving a higher social position in their own career. In this respect, they have a ‘dominant’ position. The expression was first used by Bisseret (1968), but developed more rigorously by Boudon (1977, 1984), who constructs a ‘coefficient of dominance’. This coefficient measures the ability of individuals from higher classes to take away the best social positions from individuals of the same level of education, but whose social milieu is lower.
Interestingly, the dominance structure is also characteristic of even meritocratic developed societies. Indeed, there is a lot of evidence that in modern societies a person's social status is positively correlated with his or her level of education. However, the life chances of people are still unequal since the individual's educational achievement is correlated with his or her social class. According to Boudon (1984), at each stage of the educational ladder the decisions that are made to follow a certain direction are strongly correlated with a person's class belonging. As a consequence, the social class during an individual's childhood plays a determinant role in the achievement of high socio-economic status by way of the better opportunities given to children of higher classes to get higher levels of education. Furthermore, among those who did manage to achieve the same level of education, individuals originating from higher social strata have greater probability of reaching higher social positions: the dominance effect increases the influence of the milieu, and thus raises the life chance inequalities.
There may be several different (and yet related) determinants of the dominance effect. Aside from the more general concept of inequalities of chances among people with different social origin discussed by Boudon (1973), researchers have considered socioeconomic capital (Borjas 1992), monetary transfers (Spilerman 2004) and differences in benchmarks in pursuing one's lifetime achievements (Frank 1985; Clark and Oswald 1996). Of course, the dominance effect varies by social context. Some social structures are more meritocratic than others. For instance, there is some evidence that the dominance effect is stronger in France than in the USA (Boudon 1984, p. 230).
In the context of migration, the difficulties that the children of some migrants face (school failure, precarious jobs, or even marginality) have typically been explained (among other factors) with their parents’ low level of education and low status in the labour market (Aubert et al.1997). Furthermore, there is a fast growing literature on social or ethnic capital, which explains human capital development with the quality of the environment where a person grows up or lives (e.g., Loury 1977; Borjas 1992; Lundberg and Startz 1998). The quality of the environment where a child grows up acts as an externality in the development of human capital that is beyond the control of the parents. Children raised in disadvantaged environments are ‘pushed down’ by this externality, while children raised in high skill neighbourhoods are ‘pushed up’. Social capital and human capital externalities help explain the fact that certain racial and ethnic groups do particularly well generation after generation, while other ethnic groups do poorly for a very long time (Borjas 1992).
The social capital literature on migration tends to focus on the human capital development and labour market performance of second and third generation migrants. At the same time, the obstacles that first generation migrants themselves encounter when trying to climb the social ladder in the destination country are rarely related to the influence of the social status of the parents in the country of origin. The assumption seems to be that their past was deleted when they crossed the border. Our hypothesis is that this is not the case, and the migrants’ lives in the host country are still shaped by their past social background, which is one of the main determinants of their integration success, together with other strong influences such as their level of education, and the political, economic and social conditions they face in the host country.
2.2. Empirical specification
One of the most formidable tasks in our study is the use of an appropriate indicator of the socio-economic position of the immigrants in the country of origin, which is strictly comparable to the one defining their socio-economic position in the destination country. We assume that the complexity of the social status1 can be captured through an appropriate definition of the socio-economic status of the migrants during their childhood in the country of origin and at the end of their careers in the country of destination. While a number of studies have used occupational and status ranking methods to compare the social status of parents and their children both within the same country and across different countries (Erickson and Goldthorpe 1992; Gazeboom and Treiman 1996), such methods become questionable when comparing such different socio-economic environments as those of Europe and Sub-Saharan Africa. To take an extreme example, an African village chief in the rural part of Congo may be seen as belonging to the upper class even if he is illiterate. Clearly, it will be difficult to compare that position with the position of the chief's offspring that may decide to migrate to France and live in a big city.
To overcome these obstacles, and in keeping with the literature on subjective or reference-point related evaluation of personal achievements (Frank 1985; Clark and Oswald 1996), we rely on indicators that position each respondent in the socio-economic structure of the destination country. Specifically, we use subjective status evaluation questions that define the respondent as being either rich, middle-class or poor according to the socio-economic structures of the origin and the destination country.2 First, we construct a categorical variable, which takes the value of 0 if the respondent's position is ‘poor’, 1 if the respondent's position is ‘middle-class’ and 2 if the respondent's current status is ‘rich’. We then define similar variables which define the respondent's parents as either poor, rich or middle-class in the country of origin at the time when the respondent was 15 years old.
It is obvious from the above discussion that a positive impact of a rich socio-economic status during childhood and a negative impact of a poor socio-economic status during childhood, with the omitted variable indicating a middle-class socio-economic status, on the probability of having a higher current socio-economic status would give support to our hypothesis of a dominance effect. In other words, our main hypothesis is α3>0, α4<0. In keeping with the literature, we also expect the coefficients of all other control variables to be positive (Borjas 1994; Lalonde and Topel 1997).
3. Data and descriptive statistics
Our empirical analysis is based on the Passage à la Retraite des Immigrés (PRI) data set collected by the Caisse Nationale d'Assurance Vieillesse (CNAV) and Institute National de la Statistique et des Etudes Economiques (INSEE) in Paris between December 2002 and November 2003.4 This is a representative sample of the diverse nationalities of immigrants in France at retirement age and age close to retirement. It includes very rich information on middle-aged and elderly immigrants, their parents, spouses and children, such as age, education and professional achievements, migration history, origin details, health status and wealth. The sample is restricted between ages 45 and 70 and allows us to evaluate the performance of these immigrants at the latest stages of their careers. A migrant in our case is a person born abroad to foreign parents.
To assure representativeness, the sample was constructed by way of random selection on the basis of the population census including 10,000 residences of immigrants aged 45–70 from 12 regions of the country, accounting for 90 percent of the population of immigrants in this age group in France.5 About 51 percent of the respondents come from Europe, 38 percent come from Africa, and 11 percent from all other continents. Six highly represented places of origin, Portugal, Italy and Spain from South Europe, Algeria, Morocco and Tunisia from North Africa, and all of the Sub-Saharan African migrants together account for more than 70 percent of the respondents. To assure both comparability of the origins of the migrants and sufficient sample sizes, we therefore restrict our analysis to these highly represented places of origin and perform our regressions over each individual country-based or regional sample.
Given the low level of employment among females and hence high level of dependence of their current socio-economic status on that of their spouses, we further restrict our analysis to the performance of male migrants. As indicated earlier, our main indicators of the socio-economic status of the parents of each male migrant is the subjective assessment of his living standards during childhood (age 15 of the respondent), while the migrant's current socio-economic status is captured by the subjective evaluation of his today's living standards. The two questions are identical and allow comparability of the answers.
3.1. Changes in socio-economic status
Figure 1 highlights the distribution of the socio-economic positions of migrants from different origins. The upper part of the figure highlights the distribution of socio-economic strata during the migrants’ childhoods, while the lower part of the figure highlights the present distribution of migrants’ socio-economic positions. Despite the slight differences across countries of origin, in particular the higher upward mobility among poorer migrants from European origins than migrants from African origins, this figure reveals an overall significant level of upward intergenerational mobility. The proportion of poor migrants decreases from around 20 percent to less than 5 percent among migrants of European origin and from about 20 or 30 percent to about 10 percent among the migrants of African origin. The proportion of middle class migrants increases on average by about 10 percent and reaches 70–80 percent across all countries of origin. Finally, except for migrants of Sub-Saharan Africa, for whom the proportion of rich migrants decreases, the proportion of rich individuals increases by 20–40 percent among the migrants from European origin and by about 10 percent among the migrants of North African origin.
Socio-economic positions of migrants in the countries of origin and destination.
Socio-economic positions of migrants in the countries of origin and destination.
Figure 2, which shows the proportions of migrants who move from each respective original state – poor, middle-class or rich in the country of origin to any of the outcome states – poor, middle-class or rich in the country of destination, provides further details on the upward intergenerational mobility. To give a detailed example, the proportion of poor Portuguese migrants who remain poor is 0.06 compared to the approximately 0 proportion of rich Portuguese migrants who become poor in the aftermath of migration and the proportion of 0.02 middle class migrants who become poor. At the same time, the proportion of poor Portuguese migrants who become middle class is 0.82, compared to the proportion of 0.74 middle class migrants who remain middle class and the proportion of 0.68 rich migrants who become middle class. Finally, the proportion of poor Portuguese migrants who become rich is only 0.13, compared to the proportion of 0.24 middle class migrants who become rich and the proportion of 0.32 migrants who remain rich.
With several slight differences, the mobility patterns for Italy and Spain are very similar to those of Portugal. Specifically, for the Italians we observe higher probabilities for migrants of all original socio-economic strata to become rich, lower probability of poor migrants to remain poor or become middle class and significantly higher probability of rich migrants to become middle class. Spain, on the other hand, exhibits higher probability of middle class and rich migrants to become rich and significantly lower probability of rich migrants to become middle class compared to Portugal.
The main difference in the patterns of intergenerational socio-economic mobility of the migrants from African origin is that, in general, originally poor migrants face significantly lower probability of moving upward in the social hierarchy when compared to migrants with European origin. For example, the proportion of Algerian migrants who used to be poor during childhood and remain poor is about 0.22, the proportion of Algerian migrants who used to be middle class and become poor is 0.08, and the proportion of rich Algerian migrants who become poor is 0.04, compared to the approximately 0 proportions of European migrants from any original socio-economic strata to become poor. The proportions of Tunisian migrants who become poor are very similar to those of Algerian migrants, while the proportions of poor Moroccan migrants who remain poor, the proportion of rich Moroccan migrants who become poor and the proportion of middle-class Sub-Saharan African migrants who become poor are higher than those of Algerian and Tunisian migrants.
The observation that African migrants experience higher probability of descending in the social hierarchy of their destination country than migrants from other parts of the world is puzzling. One hypothesis could be that they are more closely tied to their family and social network left in the country of origin, which gives them more constraints (obligations) towards them (sending more money, taking care of relatives or friends needing help). The strong solidarity obligations could restrict their capacity to invest in themselves and their children. In addition, we could assume that African migrants find it more difficult to improve their socio-economic situation because they are racially discriminated. It is also possible that the education and other skills acquired in Africa may have lower rentability in France. However, since African migration has so far been understudied, it is difficult to find a satisfactory explanation of this phenomenon.
Importantly, despite the cross-country differences observed, there is a clearly lower tendency for migrants of lower socio-economic strata to move upward compared to migrants of higher socio-economic strata and a lower tendency of migrants from higher socio-economic strata to move downwards compared to migrants of lower socio-economic strata. This observation is especially interesting, given the fact that the probability of acquiring a better status than that of their parents would be higher among people originating from lower socio-economic strata, as a consequence of the ceiling effect. However, we see that those belonging to well-off families face a higher probability of reaching the highest socioeconomic strata even though it is more challenging for them to overtake their parents. This provides strong preliminary evidence in favor of our hypothesis of dominance effect, which we shall explore more rigorously during our regression analysis.
3.2. Descriptive statistics
The descriptive statistics based on specification (1) provide both some explanation for the mobility patterns observed and some additional priors for the regression analysis.
These statistics, reported in Table 1, highlight the means and standard deviations of age, years of education, years in France, level of knowledge of the French language and co-habitation status of the respondents by country of origin and current socio-economic status. We observe that migrants from Sub-Saharan Africa have the highest level of education, a pattern consistent with the education policy of France vis-à-vis migrants from former French colonies (Weil 2004), followed by migrants of Italian and Tunisian origin. At the same time, migrants of Italian and Spanish origin represent the oldest cohorts of migrants, while the migrants of Sub-Saharan African origin represent the youngest cohorts of migrants. The rest of the migrant characteristics are highly comparable across countries of origin, though migrants of Sub-Saharan African origin and Italian origin exhibit slightly higher levels of mastering the French language.
| Origin | Status | Age | Yrs edu | Yrs Fr. | P. French | Couple |
|---|---|---|---|---|---|---|
| Portugal | Rich | 53.07(5.02) | 7.24(3.53) | 34.2(6.91) | 0.75(0.43) | 0.94(0.25) |
| Middle | 55.31(6.49) | 6.20(2.88) | 33.97(5.55) | 0.58(0.49) | 0.94(0.23) | |
| Poor | 55.69(6.81) | 5.81(4.50) | 33.31(7.83) | 0.38(0.50) | 0.88(0.34) | |
| Italy | Rich | 58.03(7.04) | 10.01(3.76) | 44.51(9.93) | 0.90(0.30) | 0.95(0.21) |
| Middle | 59.74(7.53) | 7.99(3.84) | 43.96(9.75) | 0.80(0.40) | 0.89(0.32) | |
| Poor | 56.00(5.61) | 4.83(2.86) | 42.5(5.75) | 0.50(0.55) | 0.33(0.52) | |
| Spain | Rich | 58.83(7.59) | 9.71(4.33) | 43.75(10.40) | 0.84(0.37) | 0.89(0.31) |
| Middle | 58.15(8.09) | 7.40(4.18) | 41.51(8.34) | 0.68(0.47) | 0.88(0.33) | |
| Poor | 57.14(8.59) | 7.57(4.72) | 40.71(5.50) | 0.71(0.49) | 0.86(0.38) | |
| Morocco | Rich | 54.42(6.29) | 9.08(5.97) | 31.32(7.02) | 0.72(0.45) | 0.99(0.11) |
| Middle | 55.42(6.63) | 5.26(5.02) | 31.57(6.38) | 0.46(0.50) | 0.96(0.21) | |
| Poor | 56.76(6.81) | 3.36(4.10) | 32.53(6.13) | 0.30(0.46) | 0.88(0.33) | |
| Algeria | Rich | 56.62(7.36) | 6.98(5.85) | 36.50(10.46) | 0.78(0.42) | 0.92(0.27) |
| Middle | 57.94(6.87) | 5.04(4.88) | 37.51(9.52) | 0.61(0.49) | 0.87(0.33) | |
| Poor | 59.77(6.19) | 3.06(4.04) | 38.44(10.05) | 0.44(0.50) | 0.72(0.45) | |
| Tunisia | Rich | 54.88(7.22) | 11.29(5.21) | 34.69(7.71) | 0.86(0.35) | 0.93(0.26) |
| Middle | 55.46(7.03) | 7.66(4.98) | 33.67(6.54) | 0.68(0.47) | 0.92(0.28) | |
| Poor | 57.78(6.30) | 5.22(5.35) | 36.39(7.36) | 0.67(0.49) | 0.66(0.49) | |
| SS Africa | Rich | 53.44(6.54) | 15.24(3.26) | 25.71(9.69) | 0.94(0.24) | 0.88(0.33) |
| Middle | 54.00(6.48) | 9.17(6.60) | 26.34(8.54) | 0.73(0.45) | 0.91(0.29) | |
| Poor | 53.85(7.96) | 8.63(6.31) | 28.15(9.96) | 0.52(0.51) | 0.74(0.45) |
| Origin | Status | Age | Yrs edu | Yrs Fr. | P. French | Couple |
|---|---|---|---|---|---|---|
| Portugal | Rich | 53.07(5.02) | 7.24(3.53) | 34.2(6.91) | 0.75(0.43) | 0.94(0.25) |
| Middle | 55.31(6.49) | 6.20(2.88) | 33.97(5.55) | 0.58(0.49) | 0.94(0.23) | |
| Poor | 55.69(6.81) | 5.81(4.50) | 33.31(7.83) | 0.38(0.50) | 0.88(0.34) | |
| Italy | Rich | 58.03(7.04) | 10.01(3.76) | 44.51(9.93) | 0.90(0.30) | 0.95(0.21) |
| Middle | 59.74(7.53) | 7.99(3.84) | 43.96(9.75) | 0.80(0.40) | 0.89(0.32) | |
| Poor | 56.00(5.61) | 4.83(2.86) | 42.5(5.75) | 0.50(0.55) | 0.33(0.52) | |
| Spain | Rich | 58.83(7.59) | 9.71(4.33) | 43.75(10.40) | 0.84(0.37) | 0.89(0.31) |
| Middle | 58.15(8.09) | 7.40(4.18) | 41.51(8.34) | 0.68(0.47) | 0.88(0.33) | |
| Poor | 57.14(8.59) | 7.57(4.72) | 40.71(5.50) | 0.71(0.49) | 0.86(0.38) | |
| Morocco | Rich | 54.42(6.29) | 9.08(5.97) | 31.32(7.02) | 0.72(0.45) | 0.99(0.11) |
| Middle | 55.42(6.63) | 5.26(5.02) | 31.57(6.38) | 0.46(0.50) | 0.96(0.21) | |
| Poor | 56.76(6.81) | 3.36(4.10) | 32.53(6.13) | 0.30(0.46) | 0.88(0.33) | |
| Algeria | Rich | 56.62(7.36) | 6.98(5.85) | 36.50(10.46) | 0.78(0.42) | 0.92(0.27) |
| Middle | 57.94(6.87) | 5.04(4.88) | 37.51(9.52) | 0.61(0.49) | 0.87(0.33) | |
| Poor | 59.77(6.19) | 3.06(4.04) | 38.44(10.05) | 0.44(0.50) | 0.72(0.45) | |
| Tunisia | Rich | 54.88(7.22) | 11.29(5.21) | 34.69(7.71) | 0.86(0.35) | 0.93(0.26) |
| Middle | 55.46(7.03) | 7.66(4.98) | 33.67(6.54) | 0.68(0.47) | 0.92(0.28) | |
| Poor | 57.78(6.30) | 5.22(5.35) | 36.39(7.36) | 0.67(0.49) | 0.66(0.49) | |
| SS Africa | Rich | 53.44(6.54) | 15.24(3.26) | 25.71(9.69) | 0.94(0.24) | 0.88(0.33) |
| Middle | 54.00(6.48) | 9.17(6.60) | 26.34(8.54) | 0.73(0.45) | 0.91(0.29) | |
| Poor | 53.85(7.96) | 8.63(6.31) | 28.15(9.96) | 0.52(0.51) | 0.74(0.45) |
Note: The figures in brackets are standard deviations.
Despite these slight differences, we see that, on average, higher levels of education, longer periods of migration, higher level of knowledge of the language of the destination countries and co-habitation are associated with higher socio-economic strata. In other words, we once again find some support to the hypotheses outlined in section 2.
4. Regression results
The regression results based on Equation (1) are reported in Table 2. To reiterate, we regress the categorical variable of current socio-economic status of the respondents, which takes the value of 0 if the migrants is poor, 1 if the migrant is middle-class, and 2 if the migrant is rich on measures of socio-economic status of the respondent during his childhood and other stylized controls. The corresponding econometric model is an ordered probit model.
| Whole sample | Portugal | Italy | Spain | Morocco | Algeria | Tunisia | SS Africa | |
|---|---|---|---|---|---|---|---|---|
| Age | 0.0023 (0.0042) | −0.0167 (0.0107) | −0.0010 (0.0105) | 0.0169 (0.0119) | 0.0040 (0.0099) | 0.0040 (0.0099) | 0.0090 (0.0171) | 0.0166 (0.0144) |
| Yrs_edu | 0.0470† (0.0061) | 0.0207 (0.0198) | 0.0753† (0.0185) | 0.0499** (0.0194) | 0.0558† (0.0141) | 0.0348† (0.0132) | 0.0673† (0.0202) | 0.0357** (0.0167) |
| Couple | 0.4919† (0.0833) | −0.0167 (0.2364) | 0.7930† (0.2309) | 0.0378 (0.2360) | 0.8299† (0.2567) | 0.5345† (0.1533) | 0.6623** (0.2913) | 0.5041* (0.2578) |
| Rich childhood | 0.3697† (0.0802) | 0.2379 (0.2384) | 0.6819† (0.2122) | 0.7954† (0.2754) | 0.0396 (0.1890) | 0.2746 (0.1825) | 0.5934** (0.2630) | 0.4293** (0.2036) |
| Poor childhood | −0.3450† (0.0583) | −0.3455† (0.1331) | −0.0137 (0.1709) | −0.5295† (0.1866) | −0.4043† (0.1285) | −0.4793† (0.1230) | −0.1309 (0.2264) | −0.1509 (0.2348) |
| Yrs in France | 0.0055** (0.0033) | 0.0072 (0.0103) | 0.0086 (0.0074) | 0.0081 (0.0105) | 0.0043 † (0.1285) | 0.0057 (0.0063) | −0.0039 (0.0171) | −0.0026 (0.0106) |
| Perfect French | 0.2387† (0.0602) | 0.2950** (0.1343) | 0.4039** (0.1925) | 0.2651 (0.2100) | 0.1727 (0.1408) | 0.2412* (0.1248) | 0.0489 (0.2331) | 0.4046* (0.2299) |
| Portugal | −0.1842** (0.0901) | |||||||
| Spain | −0.0598 (0.0965) | |||||||
| Algeria | −0.4163† (0.0873) | |||||||
| Morocco | −0.5466† (0.0949) | |||||||
| Tunisia | −0.5081† (0.1112) | |||||||
| SS Africa | −0.8343† (0.1178) | |||||||
| Pseudo Rsq | 0.1049 | 0.0465 | 0.1034 | 0.0978 | 0.0873 | 0.0730 | 0.1044 | 0.0939 |
| _cut1 | −0.7462 (0.2727) | −2.4329 (0.6628) | −0.3826 (0.7009) | −0.2592 (0.7179) | 0.1936 (0.6007) | −0.1711 (0.5910) | 0.0854 (0.8771) | 0.7210 (0.8176) |
| _cut2 | 1.7128 (0.2748) | 0.3779 (0.6511) | 2.4507 (0.7188) | 2.4088 (0.7300) | 2.5589 (0.6097) | 2.0647 (0.5968) | 2.5233 (0.8976) | 3.1637 (0.8454) |
| N Obs | 2,661 | 520 | 394 | 283 | 501 | 538 | 204 | 221 |
| Whole sample | Portugal | Italy | Spain | Morocco | Algeria | Tunisia | SS Africa | |
|---|---|---|---|---|---|---|---|---|
| Age | 0.0023 (0.0042) | −0.0167 (0.0107) | −0.0010 (0.0105) | 0.0169 (0.0119) | 0.0040 (0.0099) | 0.0040 (0.0099) | 0.0090 (0.0171) | 0.0166 (0.0144) |
| Yrs_edu | 0.0470† (0.0061) | 0.0207 (0.0198) | 0.0753† (0.0185) | 0.0499** (0.0194) | 0.0558† (0.0141) | 0.0348† (0.0132) | 0.0673† (0.0202) | 0.0357** (0.0167) |
| Couple | 0.4919† (0.0833) | −0.0167 (0.2364) | 0.7930† (0.2309) | 0.0378 (0.2360) | 0.8299† (0.2567) | 0.5345† (0.1533) | 0.6623** (0.2913) | 0.5041* (0.2578) |
| Rich childhood | 0.3697† (0.0802) | 0.2379 (0.2384) | 0.6819† (0.2122) | 0.7954† (0.2754) | 0.0396 (0.1890) | 0.2746 (0.1825) | 0.5934** (0.2630) | 0.4293** (0.2036) |
| Poor childhood | −0.3450† (0.0583) | −0.3455† (0.1331) | −0.0137 (0.1709) | −0.5295† (0.1866) | −0.4043† (0.1285) | −0.4793† (0.1230) | −0.1309 (0.2264) | −0.1509 (0.2348) |
| Yrs in France | 0.0055** (0.0033) | 0.0072 (0.0103) | 0.0086 (0.0074) | 0.0081 (0.0105) | 0.0043 † (0.1285) | 0.0057 (0.0063) | −0.0039 (0.0171) | −0.0026 (0.0106) |
| Perfect French | 0.2387† (0.0602) | 0.2950** (0.1343) | 0.4039** (0.1925) | 0.2651 (0.2100) | 0.1727 (0.1408) | 0.2412* (0.1248) | 0.0489 (0.2331) | 0.4046* (0.2299) |
| Portugal | −0.1842** (0.0901) | |||||||
| Spain | −0.0598 (0.0965) | |||||||
| Algeria | −0.4163† (0.0873) | |||||||
| Morocco | −0.5466† (0.0949) | |||||||
| Tunisia | −0.5081† (0.1112) | |||||||
| SS Africa | −0.8343† (0.1178) | |||||||
| Pseudo Rsq | 0.1049 | 0.0465 | 0.1034 | 0.0978 | 0.0873 | 0.0730 | 0.1044 | 0.0939 |
| _cut1 | −0.7462 (0.2727) | −2.4329 (0.6628) | −0.3826 (0.7009) | −0.2592 (0.7179) | 0.1936 (0.6007) | −0.1711 (0.5910) | 0.0854 (0.8771) | 0.7210 (0.8176) |
| _cut2 | 1.7128 (0.2748) | 0.3779 (0.6511) | 2.4507 (0.7188) | 2.4088 (0.7300) | 2.5589 (0.6097) | 2.0647 (0.5968) | 2.5233 (0.8976) | 3.1637 (0.8454) |
| N Obs | 2,661 | 520 | 394 | 283 | 501 | 538 | 204 | 221 |
Note: †, ** and * indicate significance at the 1, 5 and 10 percent levels, respectively. The figures in brackets are standard deviations.
Column 1 in Table 2 highlights the results for the whole sample of migrants, which include country dummy variables with Italy being the excluded category. Columns 2–8 highlight the results for each of the origin-country based samples of migrants. These results provide strong support to our key hypothesis of a dominance effect. Specifically, in the results for the whole sample of migrants the coefficient of the rich childhood variable is positive and significant at the 1 percent level and the coefficient of the poor childhood variable is negative and significant at the 1 percent level. In addition, both variables are highly significant and with the correct signs in the regression results for the sample of migrants of Spanish origin, while the coefficient of the rich childhood variable is positive and highly significant in the regression results for samples of Italian, Tunisian and Sub-Saharan African migrants; and the coefficient of the poor childhood variable is negative and significant in the regression results for the samples of Portuguese, Spanish, Moroccan and Algerian migrants. In sum, our regression results confirm our hypothesis and priors from the preliminary descriptive analysis that migrants from lower socio-economic strata face lower upward mobility chances and migrants from higher socio-economic strata have higher upward mobility chances.
To check whether there is a statistically significant cross-country difference in these results, we performed a Wald test of coefficient equality of the rich child and poor child variables across the different national samples. With a chi-square value of 11.74, the null hypothesis of coefficient equality was rejected at any conventional level. In other words, there is a confirmation that our main results hold irrespective of the choice of sample.
The rest of our results are consistent with the rest of our hypotheses. Specifically, higher education is invariably positively associated with higher socio-economic mobility, higher levels of knowledge of the language of the country of destination is positively correlated with higher levels of socio-economic mobility, living as a couple in general has a positive impact on higher levels of socio-economic mobility among male migrants and the length of migration has either positive or insignificant impact on socio-economic mobility.
Note that the results presented in Table 2 are based on the whole sample of migrants coming from the most represented countries of origin. This includes respondents who migrated prior to the age of 15, meaning that the economic status of their parents is the status after settling in France. These respondents account for approximately 17 percent of the sample. To ensure that the results are robust to the exclusion of these respondents, we re-estimated specification 1 only for the sample of migrants who migrated after the age of 15.6 The corresponding results, which are available upon request, indicate that the choice of sample has no influence on our main conclusions.
5. Concluding remarks
The results that we presented in this paper show that the social background of immigrants in the country of origin has a strong impact on their accomplishments in the country of destination. This is true across different ethnicities and nationalities of immigrants. This important result supports our hypothesis of dominance de milieu and challenges the standardized images of groups of migrants coming from specific countries of origin. It emphasizes the fact that the social heterogeneity of migrants coming from the same country of origin, but having different life and family histories is the major determinant of their own life course.
How does the transmission of socio-economic status take place across borders? As expected, we do find that what economists call ‘human capital’, or one's level of education, has an important influence on achievements in the destination country. However, even after controlling for the level of education and other crucial variables that are typically used in analyses on migration, we find that social status has important impact on socio-economic mobility and this impact holds irrespective of the ethic and national origin of migrants.
Our hypothesis is that this observation could be explained by three main dimensions of the socio-economic position of individuals. One is the feeling of ‘social legitimacy’, given by the status de notabilité (‘notability status’) of the family (Santelli 2001), which Bourdieu calls habitus. The second important dimension is monetary transfers between migrants and family members. Those belonging to richer families face a larger probability of receiving financial help from their parents, while migrants from poor families, non only do not receive economic help, but also they have to send financial transfers (remittances) to their parents (Attias-Donfut and Wolff 2008, 2009). This diminishes their ability to invest in their own well being. The third important dimension is the help of any kind available to the migrant from transnational and family networks, which is generally called ‘social capital’. Although most research generalizes the influence of transnational networks for people coming from the same countries, it is reasonable to assume that people from richer and more powerful families would have stronger networks in the higher socioeconomic strata of the country of destination. All these dimensions are captured by the indicator of subjective standard of living, characterizing the social status of the family in the country of origin.
Footnotes
Including economic, human and social capital as well as habitus (see section 5 of this article).
Specifically, each emigrant is asked to position himself within one of the following 5 categories: (1) very poor, (2) poor, (3) middle-income, (4) rich, and (5) very rich. He is also asked to evaluate in the same way the socio-economic status of his father at the time of the migrant's childhood. We group together positions 1 and 2 and 4 and 5, respectively, as there are very few observations in some of these categories and results among them are not significantly different.
We also attempted controlling for factors such as motives for migration, legal status upon arrival in France, religion and location (urban or rural) from which the respondent has migrated. Since they tended to be insignificant and did not improve or model's performance, we omitted them from the final specification.
The data were collected through CAPI (Computer Assited Personal Interview) face to face questionnaires. The interviews took place at the homes of the respondants (after sending a letter to make an appointment). Each interview with the questionnaire lasted about 90 minutes.
The final sample of 6,211 questionnaires represents 52 percent of the initial random sample drawn from the population census. There were 1,800 (15 percent) refusals, 1,000 (15 percent) addresses not reachable and 2,980 (25 percent) households out of the field (not aged 45 to 70 or non-migrants). To take into account the difference between the initial and the final sample, a weight (coefficient of ponderation) was applied; it was calculated by including different parameters, drawn on the data of the population census (see Attias-Donfut 2006: 345).
The great majority of migrants arrived in France as young adults, alone or with their spouse and children. The minority of migrants who came with their parents mainly arrived during their childhood. They belong to the oldest waves of migrants from Spain and Italy. This strong correlation with the country of origin, as well as the very insignificant number of observations for people who have arrived with their parents in the majority of the samples, makes the variable ‘coming with their parents’ irrelevant in our regressions.
References
Claudine Attias-Donfut is a sociologist. She is the director of the Ageing Research Department of the CNAV (National Retirement Fund of Public Social Security) in France and she is also associated to the Edgar Morin Center at EHESS (Ecole des Hautes Etudes en Sciences Sociales). Her research interests are concerned with relations and transfers between generations, family and the welfare state, sociology of the life course, transition to retirement and ageing, ageing in developing countries, European comparative studies, and sociology of immigration.
Ralitza Dimova is an economist. She has been a lecturer in Economics at Brunel University in West London since September 2005. Previously, she was part of two Marie Curie projects on intergenerational transfers at CNAV in Paris and the Max Planck Institute for Demographic Research in Germany. She works mostly in the area of population economics.